I recently just started using psst which is a Spotify GUI that's much lighter. When you right click a song and go to show similar tracks u get an array of sliders corresponding to the audio analysis/features like valence, danceability, energy, etc to tweak the recommendations.
It made a light and day difference for music discoverability for me, while the default spotify radio keeps giving me songs i skip instantly multiple times along with songs I've listened to a hundred times, doing this through the API, is 100x better. I've discovered 30 new songs that I love this past week while that number has been steadily dwindling for the past 6 months using Spotify.
I have used Spotify Audio Features API to display albums and playlist on a radar charts showing acousticness, instrumentalness, energy etc. And to make recomendations (generate playlists) for similiar music based on these charateristics.
It has been fun project but now I am glad that I have never considered making anything serious out of it.
I did this project because my impression is that Spotify had been always trying to steer me not to music that I like but to music that Spotify makes most money of. It had always been paid promotions over user's tastes in music.
And I am not on Spotify anymore for years now. Apple Music have really tasteful recommendations and music curation.
I worked on the Apple Music frontend and can't help but be pleased people are still using it and are pleased. I remember recommendations being a bug priority, but I wasn't involved with that. Spotify's recommendations aren't as bad as they used to be, but it still thinks I'm more into early 2000s emo rock than the kinds of metal I'm actually into.
It's funny, I had the opposite experience - Spotify understood my taste while Apple Music didn't. (Specifically, Apple Music pushed a lot more Hip Hop/R&B music than I was used to - this was in early 2016 mind you so things may have changed since).
The cool thing that's gone now was the "Audio Features" endpoint ( https://developer.spotify.com/documentation/web-api/referenc... ). You could easily get some important values about every song, now you would probably have to run your own analysis for every song you're interested in. That's a lot harder and slower if you don't want to preprocess every single song available.
So you're here telling people who were actually using these APIs that we're wrong to be upset, because LLMs? Awesome, super helpful, thanks
LLMs require data, as I'm sure you know. This is locking up what was previously an interesting source of data, which undermines your argument over the long term
On the API front, the endpoint that's being killed that was most interesting to me is actually their music analysis one. That was super-nerdy fun to fuck around with, I had a half-finished project on that with an old job. Totally interested in hearing of feature-parity alternatives I can run locally. I'd also thought I'd sometime get around to doing some network analysis with related artists too.
I honestly don't find Spotify's recommendations all that great. I definitely experienced a broadening (perhaps deepening) of my listening early on, but my experience has been that the recommendations are pretty shallow.
I find after throwing together a playlist with some stuff I like, it'll add a few more artists to my mental roster, then nothing. I'll get thrown around in the same loop with the same tunes and artists -- usually from the more famous albums.
I don't want to sound too much like the grouchy aging hipster that I am, but recommendations engines are just one of many ways of discovering music, and I feel like y'know, the old ways were better than just paying some company to do it for me. I'm talking here about being a regular on a local music scene, smoking weed with musicians, trading MP3s on the sneakernet.
Another thing where we just pay some money for "convenience", but are left with some hollow and empty algorithmic imitation of something we once loved.
Your LLM suggestion made me do a little sick in my mouth.
Spotify's algorithm has worked very well for me in the past.
I'm pretty sure it's not LLM based though, but rather domain specific, or possibly just a simple recommendation engine ("people who like x also often enjoy y").
No, they're absolutely garbage at it. I don't even understand the thought to use LLMs in the first place. And even if they weren't garbage the whole point of a music recommendation algorithm is surface music that wouldn't be in the training set so you need a way to recall likely matches at which point you've built a recommen engine.
> I don't even understand the thought to use LLMs in the first place.
You know how people believe whatever they read, hear, and watch even though it might not be true? Well an LLM is something people read and to get over the hurdle of whether something might be true or good, you simply embrace it and ignore that it could ever be wrong. I don’t get it either as I get upset when I find out a source is mostly wrong.
Doesn't affect existing apps with extended mode access, for which you have to apply and be approved. Gives you a higher ratelimit so you can ship to production. Plenty of people (me included) build small widgets for themselves without bothering to apply for extended.
> Most people writing small scripts were probably using an API key with development mode.
Yep, me included. Since my apps were never meant as anything more than utilities for myself, I never applied for extended access. Nevertheless, I used these tools multiple times a day, especially for sorting and filtering playlists by audio features such as energy and valence. Now, apparently, they will never work again. I'm sure there are plenty of other hobbyists in the same boat.
I stay on Spotify because it has an open source client spotify-qt.
I use Firefox on BSD which doesn't have DRM support so the web versions of Apple Music and Deezer don't work properly. On Apple it only plays the first 30 seconds of each song and I forget what the problem was with Deezer.
Also a real app is way nicer than a web interface of course. And with libspotify I can even change songs that play on my mobile and control it through home assistant.
None of the others allow third party clients or open source. Sure it's a niche reason but this is the reason I'm on Spotify and not somewhere else. I've tried other platforms for a month but it was crap.
I only listen to big artists anyway that are well compensated.
Apple Music has more live radio streams featuring artists too, and the Apple Music 1 radio features real live commentators. Whoever is at the controls for Music over at Apple is someone who really cares about music.
I got spotify a year ago because I needed an easy way to just put on some decent music from a playlist a friend sent me when I had people over. Since then I've realized that basically any song that I want to listen to is on there. Am I missing out on a better experience on some other platform? If so, which one and why is it better?
I fired Spotify when they signed Rogan to that ridiculous contract, and I switched to Apple Music. Great UX, great audio quality. Not prefect, but better than what I left behind.
How can you say that? Spotify held a moment early on where it was built upon pirated mp3s. At that time it was the easy way to listen to anything for free.
I remember the time where there was no party without constant Spotify ads running over the speaker, that's the only type of free account I know of.
Other than that my point was how incomplete it is and always was. It could be nice as additional catalogue to my music, but for me it's missing to many of my favourite songs to use it as main driver.
Edit:// in Switzerland downloading music for private use is no crime. So the initial situation was different I guess.
And they didn't start with illegal MP3s. They did have an ad-supported free tier from the start though. But it was not illegal. See https://en.m.wikipedia.org/wiki/Spotify
I think it's napster you're thinking of. That was an illegal sharing platform and now a mediocre paid service.
It might be Grooveshark that they're thinking of, it was notorious for quickly reuploading content that was taken down by DMCA requests: https://en.wikipedia.org/wiki/Grooveshark
Tidal is in a downwards spiral because they are running out of money, Apple subsidisies Apple Music with profits from other parts of the company.
The pot splitting model Spotify uses is definitely not good but the major labels are the ones with all the power, without pot splitting they wouldn't accept licencing to Spotify because they would make less money.
At every filthy corner of the music industry you'll find a very sore spot: the big 4 labels control this industry. From fucking with artists where contracts requiring artists to pay back all "marketing and fees" before any royalties are distributed, royalties split usually 80:20 or 70:30 for label:artist, forcing artists to make their songs viral before they can be released (without much marketing support from the labels, the only reason they exist).
It's a passion industry, and just like any other passion industry it's fraught with exploitation. Just look at game development, underpaid, overworked, because there's always someone else with passion to make a game.
They subsidise that from other parts of the business though.
Also they don't technically pay artists aside from the self-released ones, most artists with bigger payouts aren't self-released so Apple Music just like Spotify is filling major labels coffers more than the artists' pockets
That's all relevant on the comparison of why Apple Music can pay more than Spotify, unsure what you didn't get but willing to clarify.
Spotify siphons yet more income artists should be getting into corporate coffers and Daniel Ek's bank account.
No music lover should be using Spotify. They are notorious for driving the downward trend in streaming payments to artists.
They are arguably worse than the worst of the old Music Industry we were taught to hate in "tech disruptor culture 1.0".
Bandcamp revenue goes straight to artists, largely. I got 89 out of 99 dollars paid on a release of mine.
I'll open by saying that I've bought about 50 albums from bandcamp and qobuz this year, so broadly, I'm with you about supporting artists.
However, the whole "Spotify is terrible for artists" argument seems ill considered. Terrible compared to what? I lot of what I buy is relatively niche artists on relatively niche labels, who would never have been signed to a major and would never had had international distribution. These artists can't make a living through streaming, sure, but I don't think they could have made a living in the old world, either.
I still have a Spotify subscription - mostly for the family - but I use it to listen to albums before deciding to buy them. I'd buy a lot less if I couldn't vet it on Spotify first.
A lot of artists seem to think that they're entitled to make a living off their art, which seems to me to completely misunderstand the history of the music industry.
How has Bandcamp been after the last acquisition for the artists? We (end users) were all predicting its downfall, but so far the new owners haven't done anything especially egregious yet, other than laying off a bunch of staff.
I've still got my hand on the trigger waiting to download my entire library as lossless FLAC and jump ship, but so far it seems like it's been mostly business as usual.
The economics of the music industry were always heavily tilted to the record labels, but Spotify somehow took it even further. Their CEO is a billionnaire for what? Being an unprofitable middle-man that pays $1 to the labels for every $0.80 they get?
Did you honestly buy music so infrequently, or did I buy music more than the normal person? In high school, I'd buy an album/CD a week. That wasn't just new releases but also meant including back catalog to fill in the collection.
Are we just opposite ends of the music acquisition spectrum?
Many companies, for example, Amazon during its rise to power, will choose to not profit and instead reinvest in business growth and avoid tax. When there is profit, there is more tax; As i understand it, if all the revenue is allocated to expenses, it will benefit from large tax exemptions. It's sort of like running a for-profit entity as if it were a non-profit entity, though by choice and not mandate.
I think the intersection of people that are upset about a free recommendation API being cancelled and people who want a music platform that pays artists fairly is essentially zero.
Audio analysis is one of the easiest problems that ML can deal with. The problem is, how can you use a pretrained LLM for discovering newly-released music? And how do you train future models without a source of new data?
> Audio analysis is one of the easiest problems that ML can deal with.
Maybe, but that doesn’t tell me anything about LLMs. I’m not saying that it’s a particularly hard problem, I’m surprised that an LLM specifically would be good for this purpose.
And when one is building this kind of a music server, please support your favourite artists!
Ideally, if they have a Bandcamp or something similar, where you can directly buy their tracks and albums from them, do that. Usually this means that you can get access to high-quality FLACs and whatnot, but it will also mean that more money will go directly to the artists (usually money going to the record label and whatnot is unavoidable even with this, but there will still be fewer people in the middle).
And well, if that's not a thing, then at least try to buy the tracks from somewhere, so that they at least see some return on their efforts. Maybe physical CDs and the like. The point is just to be able to support your favourite artists!
Your suggestions are fine, but if you really want to support your favorite musicians, you should attend their concerts and buy merchandise there. They personally get far more profit from ticket sales and merch sales than from selling music directly. And of course, the concert experience is something way beyond just listening to a track from Bandcamp or a CD.
Uhm, is that actually true? How is the $10 I spend on a digital album on Bandcamp not 90%+ profit for the band? Sure, maybe the overpriced T-shirt has a bit more profit as a raw number, but realistically, I'm going to buy a band's merch the one time I see them every 3-5 years (assuming they stay together and tour for that long). If they release music more frequently, I would suspect buying a digital album is more sustainable long term.
I think this is also why you see bands like Weezer releasing more niche EPs/LPs. Heck, look at jam bands like Phish or Dave Matthews who release every single live show online as a separate album for fans to also buy to relive the experience they had at a particular show. The hardcore fans will buy the music, so it's in the band's best interest to "keep shipping" and record as much as possible.
This is a myth unfortunately. Unless you are a really big name artist - or a mod-level or above artist doing a show in your home town - the economics of live and touring for most musicians mean they are more likely to lose money than make money.
Imagine a band with four or five members doing a 20 date tour in 1000 cap venues where tickets are $40 each. Maths looks good, right? $40,000 a night! $800k for the tour, and then you can sell a bunch of merch an easily make $1 million. Great!
No.
A touring band might sell out every night of the tour but more likely it’s going to be 70-80% occupancy. So let’s call it 75%. Suddenly that $800k drops to $600k.
But then you need to pay the venue/promoter a big chunk of that. Depending on what the promoter is providing that could be as much as 40-50%
Let’s go with a conservative 40%.
You’re down to $360k now.
But you’ve still got to pay all the costs of the tour.
A 20 date tour probably means 25 days on the road, at least.
A tour bus that could fit 4 or 5 people plus tour manager (yes, you need one) and a tech/roadie/sound engineer to get the set up right in each venue (let’s say you’ve got one person who can do all of this) is going to cost $1500 a day for the vehicle. Add in mileage, which is often about $5+ per mile. So that 20 date tour with 25 days on the road, and 4000 miles (coast to coast) will cost you maybe $57.5k for the tour bus and driver and mileage. (Gas, insurance etc are covered by the per mile charges that tour bus operators charge). You’re going to need to park the tour bus during the day. That’s maybe $200 a day. More in some cities.
You’re down to $300k now.
But wait - no one has been paid yet!
The tour manager will easily cost $450 per day or more - and there will be days require for planning (“advancing”) the tour and wrap up days. So the 25 date tour might need 5 days advancing and two days post-tour admin. That’s $14400, so call it $15k.
Your technician will cost about the same. Maybe less, but you want someone who can do three things, so let’s call your manager plus tech/sound engineer $30k.
We are down to $240k now.
At this point it’s worth mentioning that the artist’s manager and billing agent commission on the “gross” - the entire amount the artist gets before costs - the $360k fee from tickets after the promoter’s share. Those commissions are typically 20% to manager and 15% to agent. So we need to deduct another $126k.
That gives $114k left.
None of the band members have been paid yet.
But, also, they need a support act for each show. If each support act gets $500 then that’s another $10k gone. $104k left.
Everyone needs a per diem! 7 people on the road, plus driver. They all need coffees, water, laundry, dry cleaning, gym passes, cough medicine, whatever, plus a “buy-out” for meals. So let’s make sure everyone has $60 a day for the buy-out and another $20 for incidentals. $16k. $88k left.
The tour - and all the gear - hasn’t been insured yet, and the band and crew don’t have insurance for medical emergencies while touring. Let’s say that’s going to cost another $3k total.
And then everyone needs flights and cabs at the end of the tour to get home. They’ll have excess luggage and instruments. So let’s call that $1500 each. Another $10k.
That’s means there’s $75k left.
The band needs to rehearse and build their live show. So that’s probably a couple of weeks rehearsal, planning, etc. So that’s a 40 day commitment.
Five people, 40 days, $75k. Each band member walks away with $15k - or $375 a day.
But how often are you going be doing a tour of that scale? Once a year probably. And touring is gruelling.
If you’re playing bigger venues with higher ticket prices there is more money - but costs can also scale.
To make $75k from bandcamp you need to sell maybe 10,000 $10 albums.
To make $75k from streaming you’d need maybe 18-20 million streams.
And you can do that without the crippling costs of touring.
Sure, if you’re on a label you’re going to get a lot less.
But touring isn’t a magic money tree, and it’s hard work.
Even the 75K is super optimistic. If you are touring you want to bring your own equipment, much easier to run a show that is roughly the same in multiple places then relying on the house rig in multiple venues. Once you factor in bringing your own gear (Almost definitely rented, sound is expensive, lighting even more so), you now need to factor in maintenance of the equipment and rigging.
Also the assumption you can get one guy to do lighting and sound is pretty unrealistic unless your show consists of a static wash throughout the entire show. Theatrical shows you can get away with it, but that's usually because the man hours are heavily front loaded into pre-production, but with live music you will need a dedicated LX Tech and a dedicated Sound Engineer.
I moonlight as a lighting technician during the evenings and weekends, mainly working in handful small local venues, there's me running lighting and the sound engineer doing his thing. The bands playing are easily spending £300 a night just on 2 people (And this is a small venue probably about 200 cap in the main hall), youd be spending much more for a touring crew
> As we continue to review the experience provided on Spotify for Developers, we've decided to roll out a number of measures with the aim of creating a more secure platform.
I'm sorry but more secure platform to what extent exactly?
They're breaking tooling because someone might know what I'm listening to? This is so frustrating along with getting a Spotify update almost every morning.
The official app doesn't even have a way to hide all podcasts for good, you have to click "music" at the top every single time you use the app, and now this, never-ending enshitifficattion.
I think you are talking about "Get Featured Playlists", which is more geared towards Spotify created playlists, which is under the 'Browse' tab in Spotify.
> Third party integrations continue to play an important role in the way users can experience the Spotify experience through third party apps. We evaluate the set up of our platform on an ongoing basis and remain committed to ensuring it provides the best possible opportunities for developers, artists, creators and listeners.
Has the same ring as "we value your privacy. That’s why we and our 739 partners want to track everything you do, link it to your real ID and sell it off to anyone willing to buy."
Spotify announced they have shut down several API endpoints, effective immediately. They have grandfathered in existing apps that have extended mode Web API access.
I recently just started using psst which is a Spotify GUI that's much lighter. When you right click a song and go to show similar tracks u get an array of sliders corresponding to the audio analysis/features like valence, danceability, energy, etc to tweak the recommendations.
It made a light and day difference for music discoverability for me, while the default spotify radio keeps giving me songs i skip instantly multiple times along with songs I've listened to a hundred times, doing this through the API, is 100x better. I've discovered 30 new songs that I love this past week while that number has been steadily dwindling for the past 6 months using Spotify.
I've always wanted a slider that gives increasingly eclectic and random selections for the radio playlists.
Psst sounds good, I'll try it, hopefully the API changes have not affected it.
I have used Spotify Audio Features API to display albums and playlist on a radar charts showing acousticness, instrumentalness, energy etc. And to make recomendations (generate playlists) for similiar music based on these charateristics.
It has been fun project but now I am glad that I have never considered making anything serious out of it.
I did this project because my impression is that Spotify had been always trying to steer me not to music that I like but to music that Spotify makes most money of. It had always been paid promotions over user's tastes in music.
And I am not on Spotify anymore for years now. Apple Music have really tasteful recommendations and music curation.
I worked on the Apple Music frontend and can't help but be pleased people are still using it and are pleased. I remember recommendations being a bug priority, but I wasn't involved with that. Spotify's recommendations aren't as bad as they used to be, but it still thinks I'm more into early 2000s emo rock than the kinds of metal I'm actually into.
I use Apple Music over Spotify whenever possible. The Spotify UX has always been, to me, inferior. Thank you for your work!
It's funny, I had the opposite experience - Spotify understood my taste while Apple Music didn't. (Specifically, Apple Music pushed a lot more Hip Hop/R&B music than I was used to - this was in early 2016 mind you so things may have changed since).
This is very sad, and another nudge away from Spotify for me.
I remember the API being a motivator for signing up, and I've hacked together a few toys with it over the years.
Realistically now, the only benefit Spotify provides over my MP3 collection is that it's better organised.
Musicbrainz Picard + PlexAmp is a pretty good solution.
Picard sets _all_ the metadata on the music and PlexAmp uses it to create playlists with the OpenAI API.
Nice, thanks for the rec. I'd seen beets as well which also looks good.
Spotify is definitely the lowest hanging fruit for culling on my subscription list. The API was very much part of the value proposition.
I don't believe it's even that good a deal for artists. I heard the mighty Snoop Dogg makes like USD 40K a year off it or something stupid like that.
Why?
Iiuc this is just about APIs for the recommendation engine
It’s never been easier to generate recommendations (eg via LLMs and other routes)
The core functionality otherwise remains unchanged in the API
The cool thing that's gone now was the "Audio Features" endpoint ( https://developer.spotify.com/documentation/web-api/referenc... ). You could easily get some important values about every song, now you would probably have to run your own analysis for every song you're interested in. That's a lot harder and slower if you don't want to preprocess every single song available.
yup. The ability to make anything interesting with the Spotify API just got flushed.
So happy I never got started on that little dream project that’s been knocking around the back of my head for a couple years…
>The ability to make anything interesting with the Spotify API just got flushed.
I respectfully disagree but ok
So you're here telling people who were actually using these APIs that we're wrong to be upset, because LLMs? Awesome, super helpful, thanks
LLMs require data, as I'm sure you know. This is locking up what was previously an interesting source of data, which undermines your argument over the long term
Slippery slope fallacy
On the API front, the endpoint that's being killed that was most interesting to me is actually their music analysis one. That was super-nerdy fun to fuck around with, I had a half-finished project on that with an old job. Totally interested in hearing of feature-parity alternatives I can run locally. I'd also thought I'd sometime get around to doing some network analysis with related artists too.
I honestly don't find Spotify's recommendations all that great. I definitely experienced a broadening (perhaps deepening) of my listening early on, but my experience has been that the recommendations are pretty shallow.
I find after throwing together a playlist with some stuff I like, it'll add a few more artists to my mental roster, then nothing. I'll get thrown around in the same loop with the same tunes and artists -- usually from the more famous albums.
I don't want to sound too much like the grouchy aging hipster that I am, but recommendations engines are just one of many ways of discovering music, and I feel like y'know, the old ways were better than just paying some company to do it for me. I'm talking here about being a regular on a local music scene, smoking weed with musicians, trading MP3s on the sneakernet.
Another thing where we just pay some money for "convenience", but are left with some hollow and empty algorithmic imitation of something we once loved.
Your LLM suggestion made me do a little sick in my mouth.
Cool Good to know, I won’t mention it to you again
Maybe you have ondansetron around
Are LLMs actually good at music recommendation?
> Are LLMs actually good at music recommendation?
As far as I can tell, the only thing actually any good at music recommendation is (some) humans.
Spotify's algorithm has worked very well for me in the past.
I'm pretty sure it's not LLM based though, but rather domain specific, or possibly just a simple recommendation engine ("people who like x also often enjoy y").
No, they're absolutely garbage at it. I don't even understand the thought to use LLMs in the first place. And even if they weren't garbage the whole point of a music recommendation algorithm is surface music that wouldn't be in the training set so you need a way to recall likely matches at which point you've built a recommen engine.
> I don't even understand the thought to use LLMs in the first place.
You know how people believe whatever they read, hear, and watch even though it might not be true? Well an LLM is something people read and to get over the hurdle of whether something might be true or good, you simply embrace it and ignore that it could ever be wrong. I don’t get it either as I get upset when I find out a source is mostly wrong.
Remember: FLAC files don't lose API endpoints.
"Physical books don't run out of battery" type energy.
Unfortunately, this breaks a lot of custom python programs I was using to facilitate music discovery and to generate playlists for myself.
It doesn’t affect existing apps
> Applications with existing extended mode Web API access that were relying on these endpoints remain unaffected by this change.
Doesn't affect existing apps with extended mode access, for which you have to apply and be approved. Gives you a higher ratelimit so you can ship to production. Plenty of people (me included) build small widgets for themselves without bothering to apply for extended.
> These changes will impact the following Web API applications:
- Existing apps that are still in development mode without a pending extension request
- New apps that are registered on or after today's date
Most people writing small scripts were probably using an API key with development mode.
> Most people writing small scripts were probably using an API key with development mode.
Yep, me included. Since my apps were never meant as anything more than utilities for myself, I never applied for extended access. Nevertheless, I used these tools multiple times a day, especially for sorting and filtering playlists by audio features such as energy and valence. Now, apparently, they will never work again. I'm sure there are plenty of other hobbyists in the same boat.
Any great discoveries you mightn’t have made otherwise?
Apple Music and Tidal play multiple times more to artists.
People are staying on Spotify just because of inertia and because "everyone" is there, not because it's the best at anything any more.
Tidal operates at a loss of something like $xx-xxx million a year. Who cares if it pays artists more if its not even successful.
All they are really doing is using VC money to pay artists a lil bit more to seem more enticing.
I stay on Spotify because it has an open source client spotify-qt.
I use Firefox on BSD which doesn't have DRM support so the web versions of Apple Music and Deezer don't work properly. On Apple it only plays the first 30 seconds of each song and I forget what the problem was with Deezer.
Also a real app is way nicer than a web interface of course. And with libspotify I can even change songs that play on my mobile and control it through home assistant.
None of the others allow third party clients or open source. Sure it's a niche reason but this is the reason I'm on Spotify and not somewhere else. I've tried other platforms for a month but it was crap.
I only listen to big artists anyway that are well compensated.
Apple Music has more live radio streams featuring artists too, and the Apple Music 1 radio features real live commentators. Whoever is at the controls for Music over at Apple is someone who really cares about music.
It's Zane Lowe who was a highly respected music presenter in the UK.
He's very knowledgable about music, great interviewer and seems to be in his element.
> Apple Music and Tidal play multiple times more to artists.
Seems like Apple pays 3x what Spotify pays? It can't be as simple as that though.
Also Apple Music and Tidal have high bit-rate, lossless audio.
Once you've tried it there's no turning back.
And classical music gets proper special treatment!
People have been begging for that for ages on Spotify’s forum…
Never tried Tidal. What makes it better?
I refuse to use anything Apple out of principle.
I got spotify a year ago because I needed an easy way to just put on some decent music from a playlist a friend sent me when I had people over. Since then I've realized that basically any song that I want to listen to is on there. Am I missing out on a better experience on some other platform? If so, which one and why is it better?
I fired Spotify when they signed Rogan to that ridiculous contract, and I switched to Apple Music. Great UX, great audio quality. Not prefect, but better than what I left behind.
Sound quality is better and they pay artists more. That's about it.
UI quality is a subjective thing.
It never was. The idea of using a limited catalog as the sole source of my music content is like assuming Netflix is all you need on a TV.
How can you say that? Spotify held a moment early on where it was built upon pirated mp3s. At that time it was the easy way to listen to anything for free.
I remember the time where there was no party without constant Spotify ads running over the speaker, that's the only type of free account I know of.
Other than that my point was how incomplete it is and always was. It could be nice as additional catalogue to my music, but for me it's missing to many of my favourite songs to use it as main driver.
Edit:// in Switzerland downloading music for private use is no crime. So the initial situation was different I guess.
Spotify is from Sweden, not Switzerland.
And they didn't start with illegal MP3s. They did have an ad-supported free tier from the start though. But it was not illegal. See https://en.m.wikipedia.org/wiki/Spotify
I think it's napster you're thinking of. That was an illegal sharing platform and now a mediocre paid service.
Were the claims made in 2017 shown to be false? I haven't yet read the book - how widely accessible was the beta?
https://torrentfreak.com/spotifys-beta-used-pirate-mp3-files...
https://direct.mit.edu/books/book/4136/Spotify-TeardownInsid...
It might be Grooveshark that they're thinking of, it was notorious for quickly reuploading content that was taken down by DMCA requests: https://en.wikipedia.org/wiki/Grooveshark
Citation needed.
Spotify pays 70% of their revenue to music rights holders.
Which is much less than Apple and Tidal.
What are the numbers?
Tidal is in a downwards spiral because they are running out of money, Apple subsidisies Apple Music with profits from other parts of the company.
The pot splitting model Spotify uses is definitely not good but the major labels are the ones with all the power, without pot splitting they wouldn't accept licencing to Spotify because they would make less money.
At every filthy corner of the music industry you'll find a very sore spot: the big 4 labels control this industry. From fucking with artists where contracts requiring artists to pay back all "marketing and fees" before any royalties are distributed, royalties split usually 80:20 or 70:30 for label:artist, forcing artists to make their songs viral before they can be released (without much marketing support from the labels, the only reason they exist).
It's a passion industry, and just like any other passion industry it's fraught with exploitation. Just look at game development, underpaid, overworked, because there's always someone else with passion to make a game.
> Apple subsidisies Apple Music with profits from other parts of the company.
No evidence of this and it doesn't even make sense.
EU would have a field day with it and Apple likes making money wherever it can find it.
How is any of that relevant when Apple Music pays artists much more than Spotify?
They subsidise that from other parts of the business though.
Also they don't technically pay artists aside from the self-released ones, most artists with bigger payouts aren't self-released so Apple Music just like Spotify is filling major labels coffers more than the artists' pockets
That's all relevant on the comparison of why Apple Music can pay more than Spotify, unsure what you didn't get but willing to clarify.
Spotify siphons yet more income artists should be getting into corporate coffers and Daniel Ek's bank account.
No music lover should be using Spotify. They are notorious for driving the downward trend in streaming payments to artists. They are arguably worse than the worst of the old Music Industry we were taught to hate in "tech disruptor culture 1.0".
Bandcamp revenue goes straight to artists, largely. I got 89 out of 99 dollars paid on a release of mine.
I'll open by saying that I've bought about 50 albums from bandcamp and qobuz this year, so broadly, I'm with you about supporting artists.
However, the whole "Spotify is terrible for artists" argument seems ill considered. Terrible compared to what? I lot of what I buy is relatively niche artists on relatively niche labels, who would never have been signed to a major and would never had had international distribution. These artists can't make a living through streaming, sure, but I don't think they could have made a living in the old world, either.
I still have a Spotify subscription - mostly for the family - but I use it to listen to albums before deciding to buy them. I'd buy a lot less if I couldn't vet it on Spotify first.
A lot of artists seem to think that they're entitled to make a living off their art, which seems to me to completely misunderstand the history of the music industry.
How has Bandcamp been after the last acquisition for the artists? We (end users) were all predicting its downfall, but so far the new owners haven't done anything especially egregious yet, other than laying off a bunch of staff.
I've still got my hand on the trigger waiting to download my entire library as lossless FLAC and jump ship, but so far it seems like it's been mostly business as usual.
If I had bought a CD every 2nd month for 10 years I would have had 60 albums. That is about what Spotify costs.
Spotify has been making the music field even more winner takes it all than the old status quo.
The economics of the music industry were always heavily tilted to the record labels, but Spotify somehow took it even further. Their CEO is a billionnaire for what? Being an unprofitable middle-man that pays $1 to the labels for every $0.80 they get?
It pays US$0.70 for every US$1 it gets, it's in their financial reports quite easily to see.
Did you honestly buy music so infrequently, or did I buy music more than the normal person? In high school, I'd buy an album/CD a week. That wasn't just new releases but also meant including back catalog to fill in the collection.
Are we just opposite ends of the music acquisition spectrum?
> arguably worse than the worst of the old <insert industry> we were taught to hate in "tech disruptor culture 1.0"
Always has been (meme)
Spotify pays 70% of revenue to music rights holders.
They aren't profitable.
Small bands can't have fans revenue from recordings in the same way as before, since they share pot with Tailor Swift and bot farms.
It doesn't matter if its 70 or 99.99%.
They are: https://www.statista.com/chart/26773/profitability-developme...
Maybe, for the first time ever.
Though historically when Spotify has come close to making a profit, record labels see it as an opportunity to demand more or pull out.
Many companies, for example, Amazon during its rise to power, will choose to not profit and instead reinvest in business growth and avoid tax. When there is profit, there is more tax; As i understand it, if all the revenue is allocated to expenses, it will benefit from large tax exemptions. It's sort of like running a for-profit entity as if it were a non-profit entity, though by choice and not mandate.
Selling music itself pretty much never benefits anyone in a significant way outside the top 1% of bands/whatever pop music.
This is why any touring band asks you to buy merch, they eat on the money from merch
Whatever happened to "app fairness"? Oh, right -- Fairness For Me, Not For Thee.
Good to point out and I’m with you but… fwiw…
I think the intersection of people that are upset about a free recommendation API being cancelled and people who want a music platform that pays artists fairly is essentially zero.
So yeah
> with the aim of creating a more secure platform.
What a bullshit... this is an abuse of the language.
Are any alternative APIs (non Spotify) available for the functionality being deprecated in this notice?
Not any good ones unfortunately! Great opportunity for someone to make one while other platforms are still scrapeable.
You don’t need to scrape other platforms. LLMs are already probably pretty darn good at this.
Can LLMs analyze song features (danceability, instrumentalness, speechiness, tempo)?
Audio analysis is one of the easiest problems that ML can deal with. The problem is, how can you use a pretrained LLM for discovering newly-released music? And how do you train future models without a source of new data?
> Audio analysis is one of the easiest problems that ML can deal with.
Maybe, but that doesn’t tell me anything about LLMs. I’m not saying that it’s a particularly hard problem, I’m surprised that an LLM specifically would be good for this purpose.
https://news.ycombinator.com/item?id=42266735
https://blog.metabrainz.org/2024/11/28/pissed-off-by-spotify...
Godspeed to all the OSS Adversarial Interoperability reverse engineers! APIs should be a digital human right.
De-enshittify by any means necessary!
Companies are going to continue to enshittify as long as customers refuse to leave them. There's no incentive not to.
People who want to do stuff like making custom apps or scripts that use these APIs should instead be building their own music servers.
And when one is building this kind of a music server, please support your favourite artists!
Ideally, if they have a Bandcamp or something similar, where you can directly buy their tracks and albums from them, do that. Usually this means that you can get access to high-quality FLACs and whatnot, but it will also mean that more money will go directly to the artists (usually money going to the record label and whatnot is unavoidable even with this, but there will still be fewer people in the middle).
And well, if that's not a thing, then at least try to buy the tracks from somewhere, so that they at least see some return on their efforts. Maybe physical CDs and the like. The point is just to be able to support your favourite artists!
Your suggestions are fine, but if you really want to support your favorite musicians, you should attend their concerts and buy merchandise there. They personally get far more profit from ticket sales and merch sales than from selling music directly. And of course, the concert experience is something way beyond just listening to a track from Bandcamp or a CD.
Uhm, is that actually true? How is the $10 I spend on a digital album on Bandcamp not 90%+ profit for the band? Sure, maybe the overpriced T-shirt has a bit more profit as a raw number, but realistically, I'm going to buy a band's merch the one time I see them every 3-5 years (assuming they stay together and tour for that long). If they release music more frequently, I would suspect buying a digital album is more sustainable long term.
I think this is also why you see bands like Weezer releasing more niche EPs/LPs. Heck, look at jam bands like Phish or Dave Matthews who release every single live show online as a separate album for fans to also buy to relive the experience they had at a particular show. The hardcore fans will buy the music, so it's in the band's best interest to "keep shipping" and record as much as possible.
This is a myth unfortunately. Unless you are a really big name artist - or a mod-level or above artist doing a show in your home town - the economics of live and touring for most musicians mean they are more likely to lose money than make money.
Imagine a band with four or five members doing a 20 date tour in 1000 cap venues where tickets are $40 each. Maths looks good, right? $40,000 a night! $800k for the tour, and then you can sell a bunch of merch an easily make $1 million. Great!
No.
A touring band might sell out every night of the tour but more likely it’s going to be 70-80% occupancy. So let’s call it 75%. Suddenly that $800k drops to $600k.
But then you need to pay the venue/promoter a big chunk of that. Depending on what the promoter is providing that could be as much as 40-50%
Let’s go with a conservative 40%.
You’re down to $360k now.
But you’ve still got to pay all the costs of the tour.
A 20 date tour probably means 25 days on the road, at least.
A tour bus that could fit 4 or 5 people plus tour manager (yes, you need one) and a tech/roadie/sound engineer to get the set up right in each venue (let’s say you’ve got one person who can do all of this) is going to cost $1500 a day for the vehicle. Add in mileage, which is often about $5+ per mile. So that 20 date tour with 25 days on the road, and 4000 miles (coast to coast) will cost you maybe $57.5k for the tour bus and driver and mileage. (Gas, insurance etc are covered by the per mile charges that tour bus operators charge). You’re going to need to park the tour bus during the day. That’s maybe $200 a day. More in some cities.
You’re down to $300k now.
But wait - no one has been paid yet!
The tour manager will easily cost $450 per day or more - and there will be days require for planning (“advancing”) the tour and wrap up days. So the 25 date tour might need 5 days advancing and two days post-tour admin. That’s $14400, so call it $15k.
Your technician will cost about the same. Maybe less, but you want someone who can do three things, so let’s call your manager plus tech/sound engineer $30k.
We are down to $240k now.
At this point it’s worth mentioning that the artist’s manager and billing agent commission on the “gross” - the entire amount the artist gets before costs - the $360k fee from tickets after the promoter’s share. Those commissions are typically 20% to manager and 15% to agent. So we need to deduct another $126k.
That gives $114k left.
None of the band members have been paid yet.
But, also, they need a support act for each show. If each support act gets $500 then that’s another $10k gone. $104k left.
Everyone needs a per diem! 7 people on the road, plus driver. They all need coffees, water, laundry, dry cleaning, gym passes, cough medicine, whatever, plus a “buy-out” for meals. So let’s make sure everyone has $60 a day for the buy-out and another $20 for incidentals. $16k. $88k left.
The tour - and all the gear - hasn’t been insured yet, and the band and crew don’t have insurance for medical emergencies while touring. Let’s say that’s going to cost another $3k total.
And then everyone needs flights and cabs at the end of the tour to get home. They’ll have excess luggage and instruments. So let’s call that $1500 each. Another $10k.
That’s means there’s $75k left.
The band needs to rehearse and build their live show. So that’s probably a couple of weeks rehearsal, planning, etc. So that’s a 40 day commitment.
Five people, 40 days, $75k. Each band member walks away with $15k - or $375 a day.
But how often are you going be doing a tour of that scale? Once a year probably. And touring is gruelling.
If you’re playing bigger venues with higher ticket prices there is more money - but costs can also scale.
To make $75k from bandcamp you need to sell maybe 10,000 $10 albums.
To make $75k from streaming you’d need maybe 18-20 million streams.
And you can do that without the crippling costs of touring.
Sure, if you’re on a label you’re going to get a lot less.
But touring isn’t a magic money tree, and it’s hard work.
Even the 75K is super optimistic. If you are touring you want to bring your own equipment, much easier to run a show that is roughly the same in multiple places then relying on the house rig in multiple venues. Once you factor in bringing your own gear (Almost definitely rented, sound is expensive, lighting even more so), you now need to factor in maintenance of the equipment and rigging.
Also the assumption you can get one guy to do lighting and sound is pretty unrealistic unless your show consists of a static wash throughout the entire show. Theatrical shows you can get away with it, but that's usually because the man hours are heavily front loaded into pre-production, but with live music you will need a dedicated LX Tech and a dedicated Sound Engineer.
I moonlight as a lighting technician during the evenings and weekends, mainly working in handful small local venues, there's me running lighting and the sound engineer doing his thing. The bands playing are easily spending £300 a night just on 2 people (And this is a small venue probably about 200 cap in the main hall), youd be spending much more for a touring crew
Counter point, from a small touring punk band, Direct Hit!: You Don't Have to Lose Money on Tour https://www.vice.com/en/article/you-dont-have-to-lose-money-...
Have you ever heard of Direct Hit!? I think they squarely fit in the mid-level or below.
Of course that depends somewhat on the venue and the cut they take of ticket and merch sales too.
PlexAMP is awesome
>Companies are going to continue to enshittify as long as customers refuse to leave them.
True but—
Also going to remain the case as long as customers refuse to pay for things they appreciate
> As we continue to review the experience provided on Spotify for Developers, we've decided to roll out a number of measures with the aim of creating a more secure platform.
I'm sorry but more secure platform to what extent exactly?
They're breaking tooling because someone might know what I'm listening to? This is so frustrating along with getting a Spotify update almost every morning.
The official app doesn't even have a way to hide all podcasts for good, you have to click "music" at the top every single time you use the app, and now this, never-ending enshitifficattion.
Long time ago, spotify allowed us to create, modify our playlists through end points. Now, it's impossible.
When did this stop being supported by Spotify?
https://developer.spotify.com/documentation/web-api/referenc...
https://developer.spotify.com/documentation/web-api/referenc...
https://developer.spotify.com/documentation/web-api/referenc...
I think you are talking about "Get Featured Playlists", which is more geared towards Spotify created playlists, which is under the 'Browse' tab in Spotify.
> Kills API
> Third party integrations continue to play an important role in the way users can experience the Spotify experience through third party apps. We evaluate the set up of our platform on an ongoing basis and remain committed to ensuring it provides the best possible opportunities for developers, artists, creators and listeners.
Read that as: Hell yeah, we're gonna enshittify.
Has the same ring as "we value your privacy. That’s why we and our 739 partners want to track everything you do, link it to your real ID and sell it off to anyone willing to buy."
Spotify announced they have shut down several API endpoints, effective immediately. They have grandfathered in existing apps that have extended mode Web API access.